Share with your CFO
FinOps, the discipline of governing cloud spending so engineers don’t accidentally burn through budgets the way a college student burns through a first credit card, is hitting a wall. Amit Kinha, field CTO at cloud financial management platform DoiT, argues the entire function needs to shift from cost visibility to value accountability. With CFOs now staring down $50 million annual AI tool budgets and no agreed method for measuring returns, dashboards that show what was spent tell less than half the story that boards are now demanding.
What this means for your business
The CFOs who funded AI experimentation in 2023 and 2024 with a “let’s see what happens” posture are now entering renewal cycles with nothing defensible to show the board. Kinha’s framing cuts directly at that problem. Productivity gains that can’t be quantified aren’t gains; they’re faith, and faith doesn’t survive a budget review. Whether this story is about you depends on one question: when your AI vendor contract comes up for renewal, do you have a measurement system that connects usage to a revenue or cost outcome, or do you have a dashboard?
The deeper structural problem is that FinOps as currently practiced was designed for infrastructure cost, where the unit economics are relatively clean. A compute instance costs X, serves Y requests, and you can optimize the ratio. AI workloads break that model because the value isn’t in the compute, it’s in the output, and output quality is contextual. Kinha is essentially calling for a new accounting primitive, one that links model inference spending to business outcomes like customer retention, conversion, or employee throughput. That primitive doesn’t yet exist in standard tooling, which means any vendor claiming to solve it today is selling aspiration ahead of capability.
The CFO who waits for the tooling to mature before building the measurement framework internally will be two budget cycles behind the CFO who starts defining success metrics now, even imperfect ones. A falsification condition worth tracking: if major cloud providers begin surfacing outcome-correlated cost attribution natively inside their billing consoles within the next 18 months, the window for third-party FinOps platforms to own this category narrows fast, and any multi-year contract signed today deserves a closer look at the exit terms.
Based on reporting from DoiT’s Amit Kinha: FinOps needs a new ROI language for AI, originally published 2026-03-05 03:00:00.
